基于AOS格式的扩展Mumford-Shah模型图像分割算法  

An Extended Image Segmentation Algorithm of Mumford-Shah Model Based on AOS Scheme

作　　者：宋锦萍[1] 职占江[1] 刘玉霞[1] 

机构地区：[1]河南大学数学与信息科学学院,河南开封475004

出　　处：《河南大学学报（自然科学版）》2014年第2期127-130,共4页Journal of Henan University:Natural Science

基　　金：国家自然科学基金资助项目(11226168)

摘　　要：在Chan-Vese模型的基础上,给出一种基于AOS格式的扩展Mumford-Shah模型图像分割算法.扩展的Mumford-Shah模型将图像分解为几何和纹理两部分,实现图像分割的同时有效避免了纹理和噪声的影响.数值求解采用AOS格式,不仅扩大了时间步长的调节范围,还解决了参数难以选择的问题,提高了算法稳定性.实验结果表明,本文算法分割效果良好,运算效率有了较大提高.Founded on the Chan-Vese Model, this paper presents an extended image segmentation algorithm of Mumford-Shah based on AOS scheme. The extended model divides image into geometry and texture part with the realization of a texture and noise free image segmentation. With the application of AOS scheme, the numerical solution enlarges the adjusting range of time step, and also works out the difficulty in parameter decision, finally enhances the stability of algorithm. The experiment manifests that the segmentation effect of the algorithm is good, and the efficiency is highly improved, which proves to be a highly practical significance.

关 键 词：加性算子分裂 MUMFORD-SHAH模型 水平集 图像分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]